The present invention relates to a method of recording and reproducing information in and from an optical disk such as a DVD, and more particularly to run-length limited codes suitable for high integration.
For a conventional optical disk drive, user data is changed to RLL (run-length limited) codes which are NRZI modulated to be converted into two-value data of xe2x80x9c1xe2x80x9d and xe2x80x9cxe2x88x921xe2x80x9d. This data of xe2x80x9c1xe2x80x9d and xe2x80x9cxe2x88x921xe2x80x9d is assigned a mark and a non-mark which are then written in a medium.
Famous run-length limited codes for optical disks are EFM codes (U.S. Pat. No. 4,501,000) for DVD and EFMPlus codes (U.S. Pat. No. 5,696,505) for DVD. EFM and EFMPlus codes are (d, k) RLL (run-length limited) codes where d=2 and k=10. The minimum value of the number of 0""s inserted between 1 and 1 of the run-length limited codes is represented by d=2, whereas the maximum value thereof is represented by k=10. With the NRZI modulation, the sign of a run-length limited code xe2x80x9c1xe2x80x9d is inverted as 1xe2x86x92xe2x88x921 and xe2x88x921xe2x86x921, whereas the sign of a run-length limited code xe2x80x9c0xe2x80x9d is remained unchanged. Therefore, d+1 is the length of a shortest mark or non-mark. Namely, the length of the shortest mark or non-mark of the EFM code or EFMPlus code is 3 bits.
Under such background, if the length d of the shortest mark is shortened, a conversion ratio m:n of user data to run-length limited codes becomes small, which broadens a detection window and is advantageous in terms of jitters. However, since the shortest mark of an optical disk is set to have a shape like a circle, as the length of the mark becomes short, the width of the mark becomes narrow. It arises therefore the problem that a signal amplitude corresponding to the minimum size of a mark becomes low in proportion to the square of the mark length. For example, if user data is written at the same density, a (1, 7) code of d=1 has a conversion ratio of 2:3 and a (2, 7) code of d=2 has a conversion ratio of 1:2. However, since (1, 7) code : (2, 7) code=8:9, the (1, 7) code is shorter by 8/9 than the (2, 7) code. Since the signal amplitude is proportional to the square of the mark length, (1, 7) code : (2, 7) code=64:81 so the signal amplitude of the (1, 7) code becomes about 3/4 of that of the (2, 7) code. As the signal amplitude becomes small, an S/N ratio lowers so that errors are likely to occur, which poses the problem that it is impossible to lower the conversion ratio by reducing d and broaden the detection window.
In order to solve this problem, this invention provides modulation codes in which the shortest length of the mark and the shortest length of the non-mark are set asymmetrically and the length of the mark is set to 3 bits or longer.
A comparison table between modulation codes of the invention and conventional modulation codes is shown in Table 1.
If the user bit time T is made equal, although conventional (1, 7) modulation codes have a detection window width Tw broader than (2, 10) modulation codes, there is the problem that the shortest mark length (shortest gap length) is short and a sufficient amplitude cannot be obtained. On the other hand, (2, 10) modulation codes have a shortest mark length (shortest gap length) longer than (1, 7) modulation codes, and the detection window width Tw is narrower.
Asymmetric codes of the invention have the shortest mark length longer than conventional (1, 7) and (2, 10) modulation codes and an intermediate detection window width between both modulation codes. However, since the shortest gap is shorter than (1, 7) modulation codes, modulation codes of the invention is associated with the problem that if a conventional method of directly slicing a detected waveform is used, a sufficient amplitude cannot be obtained from the shortest non-mark and the edge cannot be detected. An average interval of shortest marks (gaps) is approximately that of conventional modulation codes, as shown in Table 1. The average interval of shortest marks of asymmetric codes, particularly (8-14) asymmetric codes, is approximately that of (2, 10) modulation codes having a longest interval and the detection window width is broadened. From these re a sons, the invention provides an approach to enabling to write the shortest mark shorter and the shortest gap longer and detect asymmetric codes.
FIG. 1 shows the position relation between shortest marks and gaps of a combination pattern written in a disk.
In FIG. 1, modulation codes obtained by converting n user bits into m channel bits are represented by n-m modulation codes. FIG. 1 shows the relation between a user bit interval T and each channel bit interval, by taking as examples conventional 8-16 modulation codes and 2-3 modulation codes (which are (1, 7) modulation codes presently used widely) and 8-15 asymmetric codes and 8-14 asymmetric codes. In the upper area of asymmetric codes, the layout of marks formed by writing modulation codes themselves is shown, and in the lower area, marks before write compensation are shown by broken lines and marks after write compensation are shown by solid lines. According to the invention, when data of asymmetric codes is written, all the marks are recorded shorter, for example, as shown in FIG. 1, by about 0.5 Tw in order to make the shortest marks and gaps have an equal interval. In this manner, even a combination of the shortest mark and gap forms a write mark and a write gap each having a length of 2.5 Tw. A sufficient signal amplitude for the shortest gap can be obtained similar to conventional (2, 10) modulation codes so that front and rear edges can be detected reliably. Since the detection window width is broader than conventional (2, 10) modulation codes, the reliability of data detection can be improved. As compared to conventional (1, 7) modulation codes, although the detection window width is not broader, the reproduction signal amplitudes for the shortest mark and gap are high so that the reliability of data signal detection is excellent.
A first reproduction method is to separately detect the positions of front and rear edges of a mark. With reference to FIG. 4, a read head in relative motion along a track reads an analog signal corresponding to data recorded in a track 1. A quantized signal such as indicated in (c) is obtained from the read analog signal by using a proper threshold value 4. The analog signal indicated in (b) has an amplitude, waveform and the like which are considerably changed in accordance with the read/write conditions. It is known that at least the rising part and falling part of this analog signal waveform are represented by the same form (function). It is also know that the interval of rising parts and the interval of falling parts are equal to the intervals of write data. Namely, the relation between the digitalization threshold value and the obtained digital signal is reflected upon the function of the waveform and therefore is regular. By utilizing this fact, signals corresponding to the front and rear edges of write data are separately detected. A timing signal can be reproduced from each of the detected signals and data can be reproduced by using the timing signal. By incorporating this first method into the invention, the above-described problem can be solved.
A second reproduction method is realized by using the first reproduction method and a partial response maximum likelihood (hereinafter abbreviated to PRML) modulation method. A partial response (PR) is a method of reproducing data while a necessary signal bandwidth is compressed by positively utilizing intersymbol interference (interference between reproduced signals corresponding to adjacent written bits). A Viterbi demodulation method (ML) is one kind of maximum likelihood sequence estimation and reproduces data in accordance with the information of signal amplitudes at a plurality of time instances, while a rule of intersymbol interference between reproduced waveforms is positively utilized. The details of PRML are given in xe2x80x9cRecent Developments of Signal Processing Technology of Magnetic Disksxe2x80x9d, the Institute of Electronics, Information and Communication Engineers, Papers C-II, Vol. J75-C-II, No. 11, pp. 611-623. According to conventional PRML, since the length of the mark and gap is based upon the unit of a detection window width, the maximum likelihood sequence estimation for reproduced signals is performed at the interval of detection window widths. According to the invention, since the length of the shortest mark and gap is 2.5 times the detection window width, a conventional PRML signal process cannot be used. In this invention, therefore, the maximum likelihood sequence estimation is performed at an interval of 0.5 times the detection window width. Therefore, all marks of asymmetric codes are shortened by 0.5 times the detection window width and written so that data can be reproduced by subjecting reproduced signals from written data to the PRML process.